Artificial Intelligence (AI) in recent years has shown an unprecedentedly impressive development, tending to play a catalytic role in all aspects of life. The interest of the academic community, but also of governments, is huge in the dynamics of AI and is reflected by the truly explosive amount of investment and research that is underway. Enthusiastic opinions and statements about AI are made every day, but at the same time they also bring to the fore alarming predictions about its effects. This paper aims to describe the opportunities emerging from the use of artificial intelligence and ChatGPT to improve education, but also to identify the challenges and ethical issues that arise.

The automatic classification of X-ray detections is a necessary step in extracting astrophysical information from compiled catalogs of astrophysical sources. Classification is useful for the study of individual objects, statistics for population studies, as well as for anomaly detection, i.e., the identification of new unexplored phenomena, including transients and spectrally extreme sources. Despite the importance of this task, classification remains challenging in X-ray astronomy due to the lack of optical counterparts and representative training sets. We develop an alternative methodology that employs an unsupervised machine learning approach to provide probabilistic classes to Chandra Source Catalog sources with a limited number of labeled sources, and without ancillary information from optical and infrared catalogs. We provide a catalog of probabilistic classes for 8,756 sources, comprising a total of 14,507 detections, and demonstrate the success of the method at identifying emission from young stellar objects, as well as distinguishing between small-scale and large-scale compact accretors with a significant level of confidence. We investigate the consistency between the distribution of features among classified objects and well-established astrophysical hypotheses such as the unified AGN model. This provides interpretability to the probabilistic classifier. Code and tables are available publicly through GitHub. We provide a web playground for readers to explore our final classification at https://umlcaxs-playground.streamlit.app.

We address, through the automated reasoning tools in GeoGebra Discovery, a problem from a regional phase of the Austrian Mathematics Olympiad 2023. Trying to solve this problem gives rise to four different kind of feedback: the almost instantaneous, automated solution of the proposed problem; the measure of its complexity, according to some recent proposals; the automated discovery of a generalization of the given assertion, showing that the same statement is true over more general polygons than those mentioned in the problem; and the difficulties associated to the analysis of the surprising and involved high number of degenerate cases that appear when using the LocusEquation command in this problem. In our communication we will describe and reflect on these diverse issues, enhancing its exemplar role for showing some of the advantages, problems, and current fields of development of GeoGebra Discovery.

Vehicle-sharing systems -- such as bike-, car-, or motorcycle-sharing systems -- have become increasingly popular in big cities in recent years. On the one hand, they provide a cheaper and environmentally friendlier means of transportation than private cars, and on the other hand, they satisfy the individual mobility demands of citizens better than traditional public transport systems. One of their advantages in this regard is their availability, e.g., the possibility of taking (or leaving) a vehicle almost anywhere in a city. This availability obviously depends on different strategic and operational management decisions and policies, such as the dimension of the fleet or the (re)distribution of vehicles. Agglutination problems -- where, due to usage patterns, available vehicles are concentrated in certain areas, whereas no vehicles are available in others -- are quite common in such systems, and need to be dealt with. Research has been dedicated to this problem, specifying different techniques to reduce imbalanced situations. In this paper, we present and compare strategies for recommending stations to users who wish to rent or return bikes in station-based bike-sharing systems. Our first contribution is a novel recommendation strategy based on queuing theory that recommends stations based on their utility to the user in terms of lower distance and higher probability of finding a bike or slot. Then, we go one step further, defining a strategy that recommends stations by combining the utility of a particular user with the utility of the global system, measured in terms of the improvement in the distribution of bikes and slots with respect to the expected future demand, with the aim of implicitly avoiding or alleviating balancing problems. We present several experiments to evaluate our proposal with real data from the bike sharing system BiciMAD in Madrid.

Remarkable progress has been made in recent years in the fields of vision, language, and robotics. We now have vision models capable of recognizing objects based on language queries, navigation systems that can effectively control mobile systems, and grasping models that can handle a wide range of objects. Despite these advancements, general-purpose applications of robotics still lag behind, even though they rely on these fundamental capabilities of recognition, navigation, and grasping. In this paper, we adopt a systems-first approach to develop a new Open Knowledge-based robotics framework called OK-Robot. By combining Vision-Language Models (VLMs) for object detection, navigation primitives for movement, and grasping primitives for object manipulation, OK-Robot offers a integrated solution for pick-and-drop operations without requiring any training. To evaluate its performance, we run OK-Robot in 10 real-world home environments. The results demonstrate that OK-Robot achieves a 58.5% success rate in open-ended pick-and-drop tasks, representing a new state-of-the-art in Open Vocabulary Mobile Manipulation (OVMM) with nearly 1.8x the performance of prior work. On cleaner, uncluttered environments, OK-Robot's performance increases to 82%. However, the most important insight gained from OK-Robot is the critical role of nuanced details when combining Open Knowledge systems like VLMs with robotic modules. Videos of our experiments are available on our website: https://ok-robot.github.io

A basic question within the emerging field of mechanistic interpretability is the degree to which neural networks learn the same underlying mechanisms. In other words, are neural mechanisms universal across different models? In this work, we study the universality of individual neurons across GPT2 models trained from different initial random seeds, motivated by the hypothesis that universal neurons are likely to be interpretable. In particular, we compute pairwise correlations of neuron activations over 100 million tokens for every neuron pair across five different seeds and find that 1-5\% of neurons are universal, that is, pairs of neurons which consistently activate on the same inputs. We then study these universal neurons in detail, finding that they usually have clear interpretations and taxonomize them into a small number of neuron families. We conclude by studying patterns in neuron weights to establish several universal functional roles of neurons in simple circuits: deactivating attention heads, changing the entropy of the next token distribution, and predicting the next token to (not) be within a particular set.

Cooking recipes are one of the most readily available kinds of procedural text. They consist of natural language instructions that can be challenging to interpret. In this paper, we propose a model to identify relevant information from recipes and generate a graph to represent the sequence of actions in the recipe. In contrast with other approaches, we use an unsupervised approach. We iteratively learn the graph structure and the parameters of a $\mathsf{GNN}$ encoding the texts (text-to-graph) one sequence at a time while providing the supervision by decoding the graph into text (graph-to-text) and comparing the generated text to the input. We evaluate the approach by comparing the identified entities with annotated datasets, comparing the difference between the input and output texts, and comparing our generated graphs with those generated by state of the art methods.

The success of reinforcement learning from human feedback (RLHF) in language model alignment is strongly dependent on the quality of the underlying reward model. In this paper, we present a novel approach to improve reward model quality by generating synthetic preference data, thereby augmenting the training dataset with on-policy, high-quality preference pairs. Motivated by the promising results of Best-of-N sampling strategies in language model training, we extend their application to reward model training. This results in a self-training strategy to generate preference pairs by selecting the best and worst candidates in a pool of responses to a given query. Empirically, we find that this approach improves the performance of any reward model, with an effect comparable to the addition of a similar quantity of human preference data. This work opens up new avenues of research for improving RLHF for language model alignment, by offering synthetic preference generation as a solution to reward modeling challenges.

Large language models (LLMs) have recently gained significant attention due to their unparalleled ability to perform various natural language processing tasks. These models, benefiting from their advanced natural language understanding capabilities, have demonstrated impressive zero-shot performance. However, the pre-training data utilized in LLMs is often confined to a specific corpus, resulting in inherent freshness and temporal scope limitations. Consequently, this raises concerns regarding the effectiveness of LLMs for tasks involving temporal intents. In this study, we aim to investigate the underlying limitations of general-purpose LLMs when deployed for tasks that require a temporal understanding. We pay particular attention to handling factual temporal knowledge through three popular temporal QA datasets. Specifically, we observe low performance on detailed questions about the past and, surprisingly, for rather new information. In manual and automatic testing, we find multiple temporal errors and characterize the conditions under which QA performance deteriorates. Our analysis contributes to understanding LLM limitations and offers valuable insights into developing future models that can better cater to the demands of temporally-oriented tasks. The code is available\footnote{https://github.com/jwallat/temporalblindspots}.

Accurate crop yield prediction is of utmost importance for informed decision-making in agriculture, aiding farmers, and industry stakeholders. However, this task is complex and depends on multiple factors, such as environmental conditions, soil properties, and management practices. Combining heterogeneous data views poses a fusion challenge, like identifying the view-specific contribution to the predictive task. We present a novel multi-view learning approach to predict crop yield for different crops (soybean, wheat, rapeseed) and regions (Argentina, Uruguay, and Germany). Our multi-view input data includes multi-spectral optical images from Sentinel-2 satellites and weather data as dynamic features during the crop growing season, complemented by static features like soil properties and topographic information. To effectively fuse the data, we introduce a Multi-view Gated Fusion (MVGF) model, comprising dedicated view-encoders and a Gated Unit (GU) module. The view-encoders handle the heterogeneity of data sources with varying temporal resolutions by learning a view-specific representation. These representations are adaptively fused via a weighted sum. The fusion weights are computed for each sample by the GU using a concatenation of the view-representations. The MVGF model is trained at sub-field level with 10 m resolution pixels. Our evaluations show that the MVGF outperforms conventional models on the same task, achieving the best results by incorporating all the data sources, unlike the usual fusion results in the literature. For Argentina, the MVGF model achieves an R2 value of 0.68 at sub-field yield prediction, while at field level evaluation (comparing field averages), it reaches around 0.80 across different countries. The GU module learned different weights based on the country and crop-type, aligning with the variable significance of each data source to the prediction task.